draft analyzing the influence of occupational licensing duration on

0

Draft

Analyzing the Influence of Occupational Licensing

Duration on Labor Market Outcomes*

Suyoun Han, University of Minnesota

Morris M. Kleiner, Collegio Carlo Alberto, University of Minnesota andNational Bureau of Economic Research

*We thank participants in seminars at the Labor and Employment Relations Associationmeetings and the University of Minnesota Department Of Applied Economics for comments. Wethank the Smith Richardson Foundation for their financial support of our research, but the viewsexpressed here are those of the authors and do not necessarily reflect the views or policies of theSmith Richardson Foundation.

1

Analyzing the Influence of Occupational Licensing Duration on Labor Market Outcomes

Abstract

We analyze the labor market influence of the duration of occupational licensing statutes for elevenmajor universally licensed occupations in the U.S. Time from the start of state occupationallicensing (i.e. licensing duration) may matter in influencing the labor market outcomes. Statesusually enact grandfather clauses and ratchet up requirements that protect existing workers, andincrease costs to new entrants. In addition, adding to or raising the entry barriers are likely easieronce an occupation is established in a political jurisdiction. We find that duration years ofoccupational licensure are positively associated with wages and hours worked, but negativelyrelated to participation in the labor market. Consequently, unlike other labor market publicpolicies, occupational licensing would likely influence labor market outcomes the longer it is inplace.

2

Introduction

Occupational licensure is the legal process by which governments (mostly the states in the

U.S., but also local governments and the Federal government) identify the legal qualifications

required to become licensed to practice a trade or profession, after which only licensed

practitioners are allowed by law to receive pay for doing the work in the occupation. This form of

labor market regulation has rapidly become one of the most significant factors affecting labor

markets in the United States and other industrialized countries. Over the past several decades, the

share of U.S. workers holding an occupational license has grown sharply. Estimates suggest that

over 1,100 occupations are regulated in at least one State, but fewer than 60 are regulated in all 50

States, showing substantial differences in which occupations States choose to regulate (U.S.

Executive Office of the President, 2015) To illustrate funeral attendants are licensed in nine States

and florists are licensed in only one State Estimates find that unlicensed workers earn 10 to 15

percent lower wages than licensed workers with similar levels of education, training, and

experience (Kleiner, 2006, Kleiner and Krueger, 2013). More specifically, Kleiner and Krueger

(2013) find that licensing at the State level confers a wage premium of around 17 percent, and the

combination of State and either Federal or local licensing has an estimate effect of around 25

percent. Local licenses by themselves are not associated with higher wages, and certification has

a smaller effect on wages (Gittleman, Klee and Kleiner, 2015).

Using other data and methods, the wage premium from licensing is more modest, and is

sometimes estimated as zero. For example, Gittleman, Kleiner and Klee (2015) find that workers

with a license earn around 8.4 percent higher wages on average controlling for detailed occupation,

but that licensing also confers better employment opportunities and health and pension benefits.

Moreover, being covered by a union contract provides better benefits than state licensing coverage

(Gittleman and Kleiner, 2016). Unlike the minimum wage which requires all employers that are

3

covered by the law to pay the new wage immediately, occupational licensing allows individuals

who are working in the occupation, but do not meet the current licensing requirements to continue

working. This is called “grandfathering”. In addition, the regulated occupation generally has the

ability to ratchet up the requirements with minimal constraints from policy makers (Wheelan,

1999). Again, individuals who do not meet the current requirements are allowed to keep working

with permission from the government. In our analysis we examine how time from initial licensure,

which we call duration, influences key labor market outcomes such as wages, hours worked, and

participation in the workforce.

In our examination of the influence of occupational licensing duration on the labor market

we initially review the literature of other studies of duration effects and show that ours is the first

comprehensive examination of the issue using more than one occupation. We also present evidence

that goes beyond looking at wage determination to examine hours worked and participation in the

regulated occupation. Consistent with other findings we show occupational licensing raises wages

in the regulated occupations, and that the duration of state licenses also is associated with higher

wages. In addition, we show that the duration of state licensing is associated with an increase in

the yearly hours worked by those in the occupation of between 2 and 6 percent, but that

participation in the occupation declines over time following the implementation of licensing

suggesting some limiting of labor supply. These results are consistent with a monopoly model of

regulation that results in gains to those in the occupation through higher wages and hours, but

which may restrict new entry into the occupation.

Duration in the Labor Market for Licensed Occupations

The duration of occupational statutes has been identified in other studies as a factor that

may raise wages (Timmons and Thorton and, 2008, Law and Marks , 2009). In both studies the

4

authors examined one occupation and focused on wage determination. Our study expands on these

studies by examining 11 universally licensed occupations (i.e. licensed in all states), some of which

have been regulated for over a 100 years and others that just became universally licensed during

the past decade. The number of workers in these occupations represent the vast majority of

universally licensed workers in the U.S. These occupations were chosen because the date of initial

licensure was available with the resources available to us, or there were sufficient observations in

the Census for statistical analysis. Also, the states that licensed these occupations regulated them

at different times.

The role of different institutions on wage determination and labor market outcomes

A helpful analogy of the influence of institutions in the labor market can be drawn from

unions. When unions first organize a firm or establishment the wage increases are generally small

(Freeman and Kleiner, 1990, DiNardo and Lee, 2004, and Mas and Lee, 2012). However, cross-

sectional estimates of the impact of unions is between 15 to 20 percent (Hirsch and Mcpherson,

2013). Estimates of the additional cost of having a union worker is approximately

$ 40,500 over the course of that worker’s employment with the firm (Mas and Lee, 2012).

Moreover, unions appear to raise the wages and benefits in a robust manner the longer they are in

an establishment (Freeman and Kleiner, 1990). We examine whether these outcomes are also the

case for occupational licensing.

Unions may raise wages through collective bargaining and withholding their labor services

through concerted activities to gain wages and benefits. On the other hand, occupational licensing

could raise wages by choosing the right set of regulations to restrict supply and limit tasks of

unlicensed workers, and enhance demand by signaling that they are providing a higher quality

service (Friedman, 1962, Spence, 1973). In a manner similar to unions, the institutional mechanism

5

and design that occupational licensing uses also takes time to implement and the full effects may

only reach fruition over several decades of enhancing these rules (Hurwicz, 1973).

Background on Grandfathering and Ratcheting Requirements

Initially, the influence of licensing duration on labor market outcomes was identified in an

NBER Volume published in 1945 by Milton Friedman and Simon Kuznets (Friedman and Kuznets,

1945). They noted that the American Medical Association in 1911through the implementation of

the Flexner Report ratcheted up requirements for becoming a doctor through tougher admissions

requirements, length of education in medical school, and limiting the number of new openings for

medical education (Beck, 2004). While increasing the requirements for graduation from medical

school, and pushing for tougher licensing, the Flexner Report did not require currently working

doctors to meet the same higher requirements, and this was the classic case of grandfathering (Beck,

2004). Friedman and Kuznets went on to examine the influence of the regulations more than

twenty years later in the late 1930s, and they found that doctors were able to raise their wages by

more than 17 percent more than dentists, who did not substantially change their requirements. This

is an illustration of how an occupation can raise their wages that involved rents to those who were

in the occupation and where the entry requirements to get into the occupation were raised for just

new entrants.

More recent estimates of the influence of the length of licensing statutes on wage

determination include results for massage therapists, nurses, lawyers, and barbers (Law and Marks,

2009, Pagliero, 2010, Timmons and Thornton, 2010. Timmons and Thornton 2013). The main

results suggest that for specific occupations such as massage therapists and barbers, the length of

time that a licensing statute has been in place enhances the earnings of these practitioners, but little

evidence of the influence of duration was found for nurses. However, the estimates are limited to

6

these occupations over a relatively short time period. Our estimates expand upon and provide

evidence beyond just wage determination of the effects of licensing duration on labor market

outcomes.

Although not explicitly addressed, the process occurs by allowing current practitioners to

avoid the explicit general and specific education requirements, internships, tests, and continuing

education mandates assuming that they were in “good standing” prior to the new licensing laws.

Also, it takes many years for the individuals who did not meet these requirements to leave the

occupation or retire, and as a result, the educational quality of the new entrants is higher, and they

dominate the members of the occupation only after many years. Moreover, the longer the

occupation is licensed the greater the ability of the members of the occupation to lobby the

legislature and licensing boards to ratchet up requirements for entry within the occupation for those

who might enter from unregulated states or occupations. For example, accountants increased the

years of university schooling from 4 to 5 years in the 1990s in order to attain a Certified Public

Accountants (CPA) license (Carpenter and Stephenson, 2006). In addition, physical therapist

raised their education requirements from a bachelor’s degree in the 1990s to a doctor of physical

therapy license by 2016 (Cai and Kleiner, 2015). In both cases the national professional association

promoted these enhanced requirements or ratchetting up through the state boards of licensing or

through the state legislature. Although the policies may have enhanced the educational quality of

the new workers, they could have also reduced access to the occupation by practitioners and

consumers and limited the supply of labor to the occupation.

A Licensing Model with Duration

The model uses a framework where the work of one occupation or individual cannot legally

be done without the inputs of the other occupation. The focus of the model serves as a basis to

7

inform and develop hypotheses about the empirical work, rather than as a fully specified general

equilibrium model of production of services under regulation. The model uses a modified standard

production function:

Qpt = HH = f(P(z),K)t (1)

Qnt = HL = f(P(z), N(z),K)t, (2)

where Qpt is the output produced by the licensed practitioner over time t, which we will refer to as

“high-skilled services (HH).” Qnt is the output produced by the unlicensed worker (NP) over time

t, which we will refer to as “low-skilled services (HL).” P(z) represents the licensed worker’s labor,

recognizing that output relies on their decision of personal input, and N(z) represents the NP’s

labor, recognizing that output relies on their decision of personal input. K represents the quantities

of capital inputs used in a service production function (Reinhardt, 1972).

By law, however, the technology needed for NPs to produce HL is tied to supervision of

entry by the licensed practitioner by law. Nevertheless, within a profit function, the NP wage is

tied to the decisions of the licensed practitioner to use the labor input and technology mix to the

high-skilled provider, HH. Regulation acts as a shifter of both the supply and demand curves with

long time lags for full implementation over time t due to grandfathering and ratcheting up of skills.

For example, a licensed engineer or architect can restrict the work of an unlicensed interior

designer reducing their earnings, hours worked, or the number of workers who may chose that

occupation over time. Regulated practitioners, who are generally in control of the production of

these services, can allocate relatively low-skilled work to unlicensed workers, while taking on

higher-skilled and value-added services for themselves and thus increasing their hours worked and

earnings, but still restricting the number of workers who may enter the occupation. These

theoretical issues are empirical questions that the rest of the paper examines.

8

The Rationale for Grandfathering and Ratcheting

As noted in the literature review and theory background above, the time from the passage

of occupational licensing laws may be important in analyzing regulation’s influence on the labor

market. One rationale is that states often enact “grandfather clauses” that protect existing workers

by allowing them to practice following new regulations, even though they may not meet the current

requirements, whereas new entrants must have higher entry standards than the existing members

of the occupation. A model of grandfathering is presented by Shavell where he assumes that if the

best standard in period 1 exceeds the level of risk that would be appropriate for the expected harm, and

in this case grandfathering may be desirable. If in period 2 the known harm is below a threshold,

grandfathering is optimal – parties who engaged in the activity in period 1 can maintain their period 1

level of risk in period 2– but parties who enter the activity in period 2 should take the new

conventionally optimal precaution for the known harm, and they have certainty of the outcome in the

second period (Shavell, 2007).

For the labor market, the process of older lesser trained workers leaving the workforce or

moving to other occupations, and newer workers with higher entry requirements coming into the

job takes many years or decades to work its way through the labor market resulting in higher wages.

However, the process may limit the supply of labor in the long run by increasing entry and mobility

requirements, and allow those in the occupation to gain economic benefits by limiting employment

growth. In addition, occupations could also ratchet up the requirements for already licensed

occupations. Therefore, licensing duration, the time from the implementation of occupational

licensing legislation matters, because it may take years for the full effects of occupational licensing

to be realized in the labor market, and for the analyst to observe these changes on wages, hours,

and employment.

9

To illustrate these changes in Figure 3 we show the theoretical impact of grandfathering and

ratcheting up requirements over three periods. The first period, when there is the initial

implementation of licensing, the second is the time when unregulated workers retire or change

jobs, and the third period when full implementation of the statute has occurred and there is a

management of the optimal number of individuals who are allowed to work. A similar effect of

regulation would occur when the occupation ratchets up the requirements for entry such as the

increase in accounting or physical therapy with licensing going from four to five years or the

addition of an advanced degree for licensure

The Empirical Model

In order to develop a model with a sufficient time line to analyze how duration may

influence labor market outcomes, we use all available data from the Census and the American

Community Survey (ACS) for years from 1950 to 2013 for a sixty three year time period to

measure duration effects. We begin with 1940 since that was the first year data for individuals was

available through the Census for income, employment and hours worked, and we are in the process

of implementing these estimates. We restrict the sample to individuals who worked in eleven

major universally licensed occupations which included more than 11 million workers which more

than 10 percent of the U.S. workforce. One set of our controls were individuals who worked in

occupations that were unlicensed during our period of analysis We limit our analysis to those

occupations that have sufficient observations in the Census and were licensed in all states by the

end of our period of analysis 2013. An illustration of the licensed occupations that are in our sample

and the time line, are presented in Figure 1 during the period 1950 to 2013 (Meyer and Osborne,

2005). Finally, our sample includes individuals who were either in one of the major universally

licensed occupations when they became regulated or an unlicensed occupations for the period. Our

10

analysis is limited because we can only include individuals who are coved by licensing statutes,

but some may not have attained a license (Gittleman and Kleiner, 2016). Also, we cannot cover

the same individuals over their career as in smaller data sets such as the National Longitudinal

Survey of Youth (NLSY). However, we define individuals who worked in major licensed

occupations as a treatment group after they received a license, and individuals who worked in all

other unlicensed occupations or were unlicensed prior to their state passing a licensure exam as a

comparison or control group1.

Next, we include standard human capital variables from the Census and for more recent

years the ACS, such as gender, age, education, and potential experience. In order to generate a

reliable sample for our analysis, individuals whose education are “below 12th grade without a

diploma” for dentists, lawyers, accountants, and pharmacists were dropped. In addition,

individuals whose age is greater than 65 and whose years of potential experience are estimated to

be below zero also were deleted.

Finally, hourly real earnings were determined by dividing the annual earnings by annual

hours worked, measured by the 2013 Consumer Price Index (CPI). Annual hours worked were

calculated by multiplying the usual working hours by the number of weeks for the past 12 months.

We eliminated from the sample those with more than 60 hours of work in a week. In addition to

these restrictions, the original sample was trimmed by excluding individuals with wages below the

federal minimum wage level in that year. The resulting sample consists of 994,142,141

observations from 1950 to 2013 using the Census and the ACS sample. In Table one we show the

1 We also estimate a propensity matching approach using the nearest neighbor method to implement furtherrobustness checks of the estimates we provide in our analysis in the tables and find similar results.

11

means and standard deviations of the licensed and unlicensed workers in our sample with wage

data adjusted by the 2013 consumer price index to standardize our results.

Implementing Descriptive and Causal Estimates

The following sections present both descriptive and causal estimates of our empirical

models using initially a kernel estimation approach for a descriptive approach, and then using a

difference in difference model causal model that takes into account the different times for each of

the occupations in our model that initially became licensed in each of the 50 U.S. states over the

time period of the analysis.

Nonparametric Kernel Estimation

In order to provide basic descriptive data for our analysis for new and older licensed

occupations we developed a nonparametric kernel estimation procedure. The estimates for the

effects of the occupational licensing duration on wage determination, the kernel estimation

procedure is shown in Figure 22. The estimates are a form of data smoothing. We try to visually

inspect whether the longer duration of licensing exhibits wider variance in earnings in the

distribution of earnings. The figure in the black line represent the wage distribution for the

unlicensed workers, the pink one is the distribution for the licensed workers whose licensing

duration are less than 10 years and the blue one is the distribution for the licensed workers whose

licensing durations are more than 10 years. Our non-parametric Kernel estimation results suggests

that it takes at least 10 years to fully realize the full economic effects of occupational licensing on

wage determination.

2 The kernel estimation procedure develops and uses an auto regressive approach to predict observed outcomes andis an atheoretical way to provide basic forecasts of observed phenomena.

12

Empirically Modeling Duration Effects

In order to more fully model the influence of occupational licensing on wage determination,

hours worked, and participation in the labor market we use a basic difference in difference

approach. Since each of the states implemented their new statute at different times, we are able to

develop a background of causal inference for the influence of duration on labor market outcomes.

We would expect the relationship to initially move slowly as both new more skilled workers enter,

but less skilled grandfathered workers continue in the occupation. When grandfathered workers

retire or leave the occupation, then wages would increase more rapidly. Further, when workers

more fully control the supply of labor by ratcheting up requirements, this would also result in

wages increasing. This hypothesized relationship between time of first licensure and wage

determination over time is shown in Figure 4.

To causally link occupational licensing and labor market outcomes, we employ a difference

in difference (DID) strategy using data on changes in state licensing requirements for the eleven

occupations in our sample. Such changes affect the ability of someone to work in a licensed

occupation in a particular state without needing to fulfill additional licensing requirements. For

estimation purposes, our model would take the following general form:= + + + + + +where is an indicator of a labor market outcome such as earnings, hours worked, or

participation in the occupation, is duration of an initial occupational licensing statute,

and is the DID estimate of the effect of the change on difference in difference (DID) strategy using

data on changes in state licensing requirements for the eleven occupations in our sample, is

individual characteristics (education level, male, race, potential experience), include year fixed

effects, include state fixed effects and to include the size of the occupation in the industry.

13

Such changes could affect the ability of someone to be able to work in a licensed in that state. We use

the difference-in-difference (DID) strategy exploiting changes in state licensing laws and

requirements over time in each of the tables presented in the rest of the paper. Our source of

identification are the changers in states that adopted occupational licensing laws over time relative

to the non-adopters, individuals who were licensed in the same occupation in comparison to those

who did not achieve licensure, and any individual who was licensed relative to those who were not

licensed, In order to focus just on changers we develop separate estimates for occupations that

were licensed during the 1950 to 2013 period. However, we also present estimates of all 11

occupations in our sample many of whom were initially regulated prior to either 1940 or 1950.

Moreover, since we do not assume a linear relationship between licensing adoption and their labor

market effects, consequently we also present nonlinear estimates.

In Table 2 we show the influence of duration of the passage of a licensing statute and

licensing on earnings using clustered standard errors. We show both a linear and quadratic

specification in the table. In addition, we show in Panel A estimates using all the occupations in

our sample. In panel B, we show estimates for just those occupations that changed their licensing

status over the period of our analysis. Finally panel C shows estimates for those occupations that

did not change their regulation status from 1940 or 1950 to 2013 as a basis of sensitivity analysis.

In the first column we show the influence of the duration of the passage of a licensing law on wage

determination with no occupation controls as a benchmark for our other specifications. The

estimates suggest that for every 10 years that an occupation is licensed wages increase by a

statistically significant 3 percent. Moreover, in column 4 becoming licensed raises earnings by

almost 18 percent which is at the high range of estimates shown by Kleiner and Krueger (2010 and

2013). We estimate our models using two digit occupation controls, but we do not introduce more

14

detailed occupational controls because they would result in identification taking place largely

through individuals who were in the occupation, but were not licensed in comparison those who

were regulated and licensed. Since our objective is to examine the influence of the change in laws

over various time periods, adding detailed occupation controls would not be an appropriate

strategy for identification or for policy purposes. Our estimates across the three specifications or

groups of occupations that were older regulated occupations versus those who were regulated more

recently found relatively small differences across the three specifications as a robustness test of

our estimates.

In Table 3 we present estimates of the influence of duration of occupational licensing on

hours worked per year using clustered standard errors. Using a similar approach to the one shown

in Table 2, we begin by estimating the influence of duration with no occupation controls and a

simple linear relationship. We also show estimates of the three panels for all the occupations in

our sample, the ones that experienced changes, and for those that did not experience any changes

in licensing status during our period of data analysis. We find that duration is associated with

increases in the hours of work. A simple linear interpretation of column one shows that for every

10 years an occupation is licensed it increases yearly hours of work by approximately 31 hours or

almost 2 percent. When two digit occupation controls are introduced the number of hours worked

per year is reduced to 9 hours per year when an occupation becomes licensed or about one percent.

Also, the estimates of increasing the hours worked due to becoming licensed is more than 73 hours

per year. The increase in hours worked per year could be due to greater demand for the licensed

occupation because consumers perceive those services to be of higher quality or the reduction in

the number of practitioners. As our theory suggests licensed workers also could be reallocating the

work of unlicensed practitioners to themselves over time. Our estimates are consistent with the

15

substitution effect being dominant, because workers want to work more hours at a higher wage.

Again, we show the influence of becoming licensed using the occupations that were regulated

during our period of analysis and those who were licensed during earlier periods. To the extent

that there is a growth in hours worked, it could be that there is a reduction in the number of

practitioners, we turn our attention to these issues in Table 4 that focuses on labor market

participation.

Perhaps one of the most speculated, but little researched areas of occupational licensing

focuses on the role of the regulated institution on the labor supply of regulated practitioners. In

table 4 we estimate the influence of the duration of an occupational license statute on the relative

number of practitioners in the occupation in comparison to individuals in unregulated occupations,

again using our three different ways of categorizing the occupations in our sample from all,

changers and non-changers. Without occupation controls the participation effect is positive, but

the coefficient is effectively zero. However the passage of a licensing statute is negative and having

a licensing law reduces the relative number of practitioners by about 1 percent for every 10 years

the occupation is licensed. Similarly, adding two digit occupation controls reduces participation

by .1 percent for every ten years of licensing. The overall effect of licensing is to reduce the relative

number of workers by about 8 percent for every 10 years the occupation is licensed. These declines

in the number of workers may be the reason that wages go up and the number of hours worked by

practitioners also increases as shown in Tables 2 and 3. Consistent with the theoretical section

occupational licensing statutes can provide the basis for regulated practitioners reallocating work

toward members in their own occupation, thus reducing the number of workers in the occupation.

The result could be the higher observed pay.

16

In order to provide a further robustness check on the estimates shown in Tables 2 through

4, in Figure 5 we show the slope of the hourly wage curve before and after the change to licensure.

Wages in the regulated occupations are relatively flat before the introduction of licensing, but

increase slightly immediately following regulation, perhaps due to signaling, and then go down

right after licensing perhaps as more new workers enter the occupation that have less experience.

After about 10 years, hourly wages increase as duration of licensing grows, perhaps due to workers

who were grandfathered leaving the occupation, and the ability of occupations to ratchet up the

legal requirements for entry. The statistically significant different slopes of the wage lines before

and after the change to licensure presented in Figure 5, and the similarities between the eleven

universally licensed and never licensed group shown in Table 1 suggests that the difference in

difference approach of identification has validity for our empirical model.

From a policy perspective, our estimates are consistent with the decision of the U.S.

supreme court in holding that dentists were operating as monopolists in reallocating tasks to

themselves when the State of North Carolina Board outlawed the work of “teeth whiteners” (North

Carolina Board of Dentistry v. FTC, 2014). Our results suggest that occupational licensing works

slowly over time as older less skilled workers retire or move to other occupations, and the state

boards or legislatures that regulate the professions ratchets up the requirements for entry. Our

ability to document these changes show how important labor market institutions work with

deliberate speed to enhance the work and pay arrangements for their members (Kleiner, 2015).

Conclusions

Since new occupational licensing statutes takes time to fully implement, duration should

matter in influencing labor market outcomes. For example, states often enact grandfather clauses

17

or ratchet up the requirements for entry that protect existing workers, and new entrants must have

higher entry standards than those already in the occupation. The process of older lesser educated

workers leaving, and newer workers with higher entry requirements coming into the occupation

takes time to work its way through the labor market. Our results show that wages rise and hours

worked per year increases, but that the relative number of workers in the occupation declines using

data covering a sixty-three year period. The results of our study should allow policy analysts and

makers to develop and implement more informed decisions on the rapidly growing labor market

institution of occupational licensing.

18

References

Beck, Andrew. 2004. "The Flexner Report and the Standardization of American MedicalEducation" Journal of the American Medical Association, 29(17) January. 2139-2140.

Cai, Jing and Morris M. Kleiner, 2015. “The Labor Market Consequences of Regulating SimilarOccupations: The Licensing of Occupational and Physical Therapists” University of Minnesotaand Upjohn Institute for Employment Research Working Paper, 2015.

Carpenter, Charles and E. Frank Stephenson, 2006. “The 150-hour rule as a barrier to enteringpublic accountancy” Journal of Labor Research , Volume 27, Issue 1 , pp 115-126.

DiNardo John and David Lee. 2004. “Economic Impacts of New Unionization on Private SectorEmployers: 1984-2001.” Quarterly Journal of Economics 119(4):1383-1441

Freeman, Richard, and Morris M. Kleiner. 1990. “The impact of new unionization on wages andworking conditions.” Journal of Labor Economics 8(1) Part 2: S8- S25.

Friedman, Milton. 1962. Capitalism and freedom. Chicago: University of ChicagoPressFriedman, Milton, and Simon Kuznets. 1945. Income from Independent ProfessionalPractice. New York: National Bureau of Economic Research.

Gittleman, Maury, Morris Kleiner, (2015). Wage effects of unionization and occupationallicensing coverage in the United States. Industrial and Labor Relations Review, forthcoming.

Gittleman, Maury., Mark Klee, and Morris Kleiner.2015, Analyzing the Labor Market Outcomesof Occupational Licensing, National Bureau of Economic Research Working Paper 20,169.February.

Hirsch, Barry T. and Macpherson, David A. 2013.Union Membership and Coverage Databasefrom the CPS, downloaded 10/30/2013 from http://www.unionstats.com/ .

Hurwicz, Leonid (1973). "The design of mechanisms for resource allocation". The AmericanEconomic Review, special issue: papers and proceedings of the eighty-fifth annual meeting of theAmerican Economic Association, (Richard T. Ely Lecture) (American Economic Association viaJSTOR) 63 (2): 1–30. JSTOR 1817047

Kleiner, Morris M. 2006. Licensing Occupations: Enhancing Quality or RestrictingCompetition? Kalamazoo, MI: Upjohn Institute for Employment Research.

Kleiner, Morris, 2015. Guild-Ridden Labor Markets: The Curious Case of OccupationalLicensing, Upjohn Institute, Kalamazoo, Michigan.

Kleiner, Morris M., and Alan B. Krueger. 2010. “The Prevalence and Effects of OccupationalLicensing.” British Journal of Industrial Relations 48(4): 676–687.

19

Kleiner, Morris M. and Alan B. Krueger. 2013. “Analyzing the Extent and Influence ofOccupational Licensing on the Labor Market.” Journal of Labor Economics 31(2): S173–202.

Law, Marc T., and Mindy S. Marks. 2009. “The Effects of Occupational Licensing Laws onMinorities: Evidence from the Progressive Era.” Journal of Law and Economics 52(2): 351–366.

Lee, David S., and Alexandre Mas. 2012. “Long-run impacts of unions on firms: New evidencefrom financial markets, 1961–1999.” Quarterly Journal of Economics 127(1): 333–78

Meyer, P. B., and Anastasiya M. Osborne. 2005. “Proposed Category System for 1960-2000Census Occupations,” BLS Working Papers 383.

North Carolina Board of Dental Examiners v. FTC, 2015. U.S. Supreme Court.

Pagliero, Mario. 2010. “Licensing Exam Difficulty and Entry Salaries in the US Market forLawyers.” British Journal of Industrial Relations 48(4): 726–39.

Reinhardt, U. 1972. “A Production Function for Physician Services.” Review of Economics andStatistics 54(February): 55-66.

Shavell, Steven, 2007. “On Optimal Legal Change, Past Behavior, and Grandfathering,” NBERWorking Paper No. 13563 October.

Spence, Michael, 1973. "Job Market Signaling". Quarterly Journal of Economics, The MITPress 87 (3): 355–374.

Timmons, Edward J. and Thornton, Robert J., 2010. “The Licensing of Barbers in the USA”.British Journal of Industrial Relations, Vol. 48, Issue 4, pp. 740-757, December 2010. Availableat SSRN: http://ssrn.com/abstract=1694828 or http://dx.doi.org/10.1111/j.1467-8543.2010.00811.

Timmons, Edward J., and Robert J.Thornton.2013. “Licensing One of the World’s OldestProfessions: Massage.” Journal of Law and Economics 56(2): 371-388.

U.S. Executive Office of the President, 2015. A Framework for Policymakers. The White House,Washington, D.C. p. 76.

Wheelan, Charles J. 1999. “Politics or Public Interest? An Empirical Examination ofOccupational Licensure.” Unpublished manuscript, University of Chicago.

20

Figure 1

1800 1850 1900 1950 2000

Lawyers DentistsPharmacists ArchitectsAccountants BarbersPhysical Therapists Registered NursesPractical Nurses Occupational TherapistsTeachers

Time line of when occupations became licensed

21

Figure 2

22

Figure 3 Illustration of the Effects of Grandfathering New Regulations

Duration=0(Unlicensed) Duration=t1 Duration=t2Pre-existingPractitioners

Pre-existingPractitioners

New Entrants

New Entrants

23

Figure 4: Hypothesized Functional Form of Relationship between Earnings and Duration of Licensing

wag

e

0 10 20 30Duration

24

Figure 5. Estimated Slopes of Hourly Wages before and After the Implementation of Licensing Statues*

*Slopes of the lines are statistically different after the implementation of a licensing statute.

25

Table 1 Means and Standard deviation of Licensed and Nonlicensed occupations

VARIABLES 11 Universally LicensedOccupations

Never LicensedOccupations

Age 42.04(11.34)

39.12(12.57)

Male 0.30(0.46)

0.49(0.50)

White 0.83(0.38)

0.79(0.41)

Experience 19.92(11.39)

19.97(12.69)

Total Worked Weeksper year

46.65(9. 78)

46.23(11.80)

Average WorkedHours per week

40.77(7.96)

39.84(8.21)

Total Worked Hoursper year

1,916.67(569.08)

1,867.99(636.66)

Log Hourly Earnings(2014 CPI)

3.37(0.57)

2.96(0.62)

licensure 0.94(0.24)

0.00(0.00)

Years of Duration 51.75(36.90)

0.00(0.00)

Industry Size (% ofGDP)

4.04(3.29)

6.20(4.26)

N 151,742392 842,399,749

26

Table 2 Effects of Licensing Duration on Log Hourly Earnings

Panel A. All Occupations

(1) (2) (3) (4) (5) (6)VARIABLES Log Hourly

EarningsLog Hourly

EarningsLog Hourly

EarningsLog Hourly

EarningsLog Hourly

EarningsLog Hourly

EarningsDuration 0.0030***

(0.000)0.0035***(0.000)

0.0021***(0.000)

0.0016**(0.001)

Duration Squared -0.0000**(0.000)

0.0000(0.000)

licensure 0.1784***(0.000)

0.0480*(0.0254)

Constant 0.8545***(0.0287)

0.8555***(0.0290)

0.8272***(0.0302)

0.8655***(0.0286)

0.8643***(0.0286)

0.8495***(0.0288)

Individual Covariates Yes Yes Yes Yes Yes YesIndustry Size Yes Yes Yes Yes Yes Yes2-Digit SOC No No No Yes Yes YesYear FE Yes Yes Yes Yes Yes YesState FE Yes Yes Yes Yes Yes YesObservations 14,453,630 14,453,630 14,453,630 14,453,630 14,453,630 14,453,630R-squared 0.325 0.325 0.322 0.330 0.330 0.330

Panel B. Occupations that changed their regulation status(1) (2) (3) (4) (5) (6)

VARIABLES Log HourlyEarnings

Log HourlyEarnings

Log HourlyEarnings

Log HourlyEarnings

Log HourlyEarnings

Log HourlyEarnings

Duration 0.0039***(0.000)

0.0053***(0.001)

0.0014***(0.000)

0.0005(0.001)

Duration Squared -0.0000(0.000)

0.0000(0.000)

licensure 0.1463***(0.010)

0.0363(0.025)

Constant 0.8938***(0.029)

0.9061***(0.036)

0.9131***(0.030)

0.9108***(0.030)

0.9064***(0.032)

0.9168***(0.030)


27

Panel C. Occupations that did not change regulation status

(1) (2) (3) (4) (5) (6)VARIABLES Log Hourly

EarningsLog Hourly

EarningsLog Hourly

EarningsLog Hourly

EarningsLog Hourly

EarningsLog Hourly

EarningsDuration 0.0027***

(0.000)0.0013*(0.001)

0.0014***(0.000)

0.0039***(0.001)

Duration Squared 0.0000**(0.000)

-0.0000**(0.000)

licensure 0.2565***(0.005)

-0.1330***(0.036)

Constant 0.8224***(0.029)

0.8243***(0.029)

0.8079***(0.029)

0.8336***(0.029)

0.8346***(0.029)

0.8308***(0.030)


Standard errors in parentheses*** p<0.01, ** p<0.05, * p<0.1

28

Table 3. Effects of Licensing Duration on Total Worked Hours per year


(1) (2) (3) (4) (5) (6)VARIABLES Total Worked

Hours per yearTotal WorkedHours per year

Total WorkedHours per year




Duration 0.3125***(0.053)

-1.6734**(0.307)

0.0897(0.200)

0.9248**(0.444)

Duration Squared 0.0199***(0.003)

-0.0069***(0.002)

licensure -25.4616***(5.671)

73.4839***(18.181)

Constant 755.7721***(23.451)

752.0063***(23.797)

730.0350***(23.639)

747.0595***(25.037)

748.8821***(25.049)

752.7786***(24.626)

Individual Covariates Yes Yes Yes Yes Yes YesIndustry Size Yes Yes Yes Yes Yes Yes2-Digit SOC No No No Yes Yes Yes

Year FE Yes Yes Yes Yes Yes YesState FE Yes Yes Yes Yes Yes Yes

Observations 14,453,630 14,453,630 14,453,630 14,453,630 14,453,630 14,453,630R-squared 0.147 0.148 0.147 0.150 0.150 0.150

Panel B. Occupations that Changed their Regulation Status







Duration -1.4147***(0.110)

-3.2553***(0.608)

0.5858*(0.347)

1.7950**(0.678)

Duration Squared 0.0345***(0.010)

-0.0177**(0.008)

licensure -72.3954***(8.935)

76.2312***(17.756)

Constant 705.2900***(23.753)

689.6890***(23.558)

692.1555***(24.102)

690.6337***(25.333)

696.7370***(24.887)

696.8075***(24.595)




29

Panel C. Occupations that did Not Change their Regulation Status







Duration 0.6760***(0.049)

0.8492***(0.207)

-0.0491(0.141)

0.1505(0.497)

Duration Squared -0.0016(0.002)

-0.0009(0.002)

licensure 69.7666***(5.273)

-160.5182***(48.984)

Constant 730.0293***(25.379)

729.8010***(25.402)

728.0400***(25.121)

730.2953***(25.605)

730.3774***(25.595)

730.3561***(25.695)




30

Table 4. Effects of Licensing Duration on Labor Market Participation


(1) (2) (3) (4) (5) (6)VARIABLES Participation Participation Participation Participation Participation ParticipationDuration 0.0001***

(0.000)-0.0000(0.000)

-0.0001***(0.000)

-0.0001***(0.000)

Duration Squared 0.0000**(0.000)

0.0000*(0.000)

licensure -0.0004(0.001)

-0.0082***(0.002)

Constant 0.8574***(0.006)

0.8573***(0.006)

0.8550***(0.006)

0.8557***(0.006)

0.8556***(0.006)

0.8558***(0.006)


Panel B. Occupations that Changed their Regulation Status

(1) (2) (3) (4) (5) (6)VARIABLES Participation Participation Participation Participation Participation ParticipationDuration -0.0000*

(0.000)-0.0001(0.000)

-0.0001***(0.000)

-0.0002**(0.000)

Duration Squared 0.0000(0.000)

0.0000(0.000)

licensure -0.0042***(0.001)

-0.0078***(0.002)

Constant 0.8361***(0.006)

0.8351***(0.005)

0.8350***(0.006)

0.8364***(0.006)

0.8359***(0.006)

0.8356***(0.006)


31

Panel C. Occupations that Did Not Change Their Regulation Status

(1) (2) (3) (4) (5) (6)VARIABLES Participation Participation Participation Participation Participation ParticipationDuration 0.0001***

(0.000)0.0002***

(0.000)-0.0000(0.000)

0.0001(0.000)

Duration Squared -0.0000***(0.000)

-0.0000(0.000)

licensure 0.0074***(0.000)

-0.0081(0.005)

Constant 0.8497***(0.006)

0.8495***(0.006)

0.8497***(0.006)

0.8493***(0.006)

0.8493***(0.006)

0.8493***(0.006)


draft analyzing the influence of occupational licensing duration on

Documents